Modeling and Prediction of the Impact Factor of Journals Using Open-Access Databases

Abstract

This article is motivated by the work as editor-in-chief of the Austrian Journal of Statistics and contains detailed analyses about the impact of the Austrian Journal of Statistics.
 The impact of a journal is typically expressed by journal metrics indicators. One of the important ones, the journal impact factor is calculated from the Web of Science (WoS) database by Clarivate Analytics.
 It is known that newly established journals or journals without membership in big publishers often face difficulties to be included, e.g., in the Science Citation Index (SCI) and thus they do not receive a WoS journal impact factor, as it is the case for example, for the Austrian Journal of Statistics.
 In this study, a novel approach is pursued modeling and predicting the WoS impact factor of journals using open access or partly open-access databases, like Google Scholar, ResearchGate, and Scopus. I hypothesize a functional linear dependency between citation counts in these databases and the journal impact factor. These functional relationships enable the development of a model that may allow estimating the impact factor for new, small, and independent journals not listed in SCI. However, only good results could be achieved with robust linear regression and well-chosen models.
 In addition, this study demonstrates that the WoS impact factor of SCI listed journals can be successfully estimated without using the Web of Science database and therefore the dependency of researchers and institutions to this popular database can be minimized. These results suggest that the statistical model developed here can be well applied to predict the WoS impact factor using alternative open-access databases.

Predicting the citation count and CiteScore of journals one year in advance

“…Although the study was designed for the regression of JIF values for a completed calendar year from which other bibliometric data is available as input features, it was shown that certain features have positive correlations with JIF and are useful when applied to a linear regression model. Another study has investigated the potential to estimate JIF values using machine learning approaches applied to bibliometric data from open-access databases Templ (2020). Such methods are motivated by the fact that for journals not indexed in SCI, the lack of public accessibility of the indexing database makes it impossible to calculate a journal's hypothetical JIF exactly as it would officially be done.…”

How Reliable Are Impact Factors for Ranking Water Resources Journals? An Analysis of the 70,878 Citation Records of the 2002 and 2021 Top 10 Journals for the Last Two Decades

“…A similar study was performed by Brzezinski (2014). Templ (2020) predicted the IF using data from Google Scholar, ResearchGate, and Scopus. Through a statistical analysis, Huang et al (2019) demonstrated that modifying a journal's status to "Open Access" leads to a higher IF.…”

“…A similar study was performed by Brzezinski (2014). Templ (2020) predicted the IF using data from Google Scholar, ResearchGate, and Scopus. Through a statistical analysis, Huang et al.…”